Contamination or Containment? Sub-state nationalism in Belgian political parties electoral manifestoes ( )

ECPR Conference Reykjavik 25-27 august 2011 Panel Multi-level party politics Contamination or Containment? Sub-state nationalism in Belgian political parties electoral manifestoes (1965-2010) Dave Sinardet & Niels Morsink Very first draft, all comments more than welcome! Dave Sinardet, Free University of Brussels & University of Antwerp, dave.sinardet@vub.ac.be; dave.sinardet@ua.ac.be Niels Morsink, University of Antwerp, niels.morsink@ua.ac.be 1

Introduction The debate on the role of federalism in so-called (culturally) divided societies is plagued by an unresolved paradox (see e.g.: Kymlicka, 1998; Simeon & Conway, 2001; Erk & Lawrence, 2009). On the one hand, by granting autonomy to sub state collectivities, federalism can be considered as a technique to regulate and manage conflicts and to temper demands of sub-state nationalist movements and parties aimed at further disintegration of the state. On the other hand, by granting such self-rule, federalism can also institutionalize, essentialise and legitimize sub state identities and thus provide tools to reinforce sub-state nationalist parties and movements demands. Obviously, this question on the effects of federalism is difficult to answer as such. Federal systems and the national contexts in which they are implemented can strongly differ amongst each other and therefore the presence or absence of other factors can play a crucial role. While the subject of much theoretical debate, this federal paradox has not been the subject of very much empirical research yet. While there is an extensive strand of literature on sub-state nationalist parties and while institutional arrangements have been used as an independent variable to explain the success of these parties (Brancati, 2007), the precise relation between sub-state nationalism and federalism hasn t often been explicitly focused on. We want to focus more in detail on the relation and precise dynamics between federalization and sub-state nationalism which we see as an important indicator to be able to speak of pacification or aggravation of ethno-regional conflict. If there is an increase of sub state nationalism, this would indicate that federalism has not lead to pacification. Such an increase should not only be measured through the rise and electoral success of sub state nationalist parties, but probably more importantly also through the spread of issue salience, positions and ideas of sub state nationalist parties, in the first place among other parties. We will focus in detail on the Belgian case, by analyzing the evolution of sub-state nationalism among Belgian political parties during the past five decades, during which the Belgian state went from a unitary to a federal institutional architecture, with a separatist party winning the most recent 2010 federal elections in the northern part of the country. More precisely, we will analyse electoral manifestoes of all parliamentary represented political parties in Belgium between 1965 and 2010. This will be done on three dimensions. The first is the salience of community issues(regionalisation, cultural identity, federal reform, etc). The second is the positions on these issues. The third and most important dimension is the arguments used by political parties to advocate these positions, so that we can assess whether besides maybe sharing the same position parties also share the same rationale behind this position. We will pay particular attention to the evolution of mainstream parties on these dimensions during the federalisation process. Were sub state nationalist parties able to make their concerns, positions and visions mainstream? In that case the federalisation process would not have reduced disintegrative tendencies and we would rather be able to speak of contamination than containment. Or could it be that (some) sub-state nationalist parties have become less radical and were co-opted by the others? Can we distinguish key moments and events that had an influence on this relationship? 2

The federal paradox, sub-state nationalism and manifesto research The literature on the federal paradox in divided societies does not come to a clear conclusion (Maiz, 2003 : 204). Many authors (e.g. Kymlicka, 1998; Gagnon & Karmis, 2001 : 137, 151; Simeon & Conway, 2001 : 364; Brancati, 2006; Sorens, 2004 : 740) argue that federalism, while it provides national minorities with a workable alternative to secession, also helps to make secession a more realistic alternative by reinforcing the belief that the group is able and entitled to exercise full sovereignty. ) Federalism would not be likely to remove the issue of secession from the political agenda, but rather to enhance the likeliness of secession, because by institutionalizing ethnic divisions, it legitimizes the ethnic claim and reinforces the believe of groups that they are separate peoples with inherent rights of self-government, whose participation in the larger country is conditional and revocable (Kymlicka, 1998 : 140). Other authors have argued however that federalism can harmonize relations between several ethno-cultural groups in a multinational state, reducing the desire for separation (O Leary, 2001; Anderson, 2004) and removing multi-ethnic tensions from the politics of the centre (Swenden, 2006 : 288). Of course, the debate on the federal paradox cannot be settled in a general way. Whether federalism leads to radicalization or moderation of conflicts in divided societies, also depends on other factors, such as characteristics of the federal system (centripetal vs centrifugal federalism, symmetrical vs asymmetrical federalism, consociational vs majoritarian, the number of federated entities, etc), the electoral system, the party system, socio-economical differences, the number of official languages, etc. (Horowitz, 1985; Reilly, 2004; Brancati, 2007; Keating, 1997; Swenden, 2006; Hechter, 1975). Before turning to our own contribution to this debate, based on the case for Belgium, we want to put into question some of the premises in the dominant literature on the federal paradox. First, it mostly links concepts and theories on federalism with those on ethnic conflict and secession (Erk & Anderson, 2009). This is often reductionist and less relevant for Western federal systems, where the question rather poses itself in terms of the development and success of sub-state nationalism, demands for greater autonomy and outbreak of political crises than of outright secession. Therefore, connecting the literatures on federalism and (sub-state) nationalism can prove very fruitful. The current literature also tends to see federal systems as static, while they and the context in which they are situated can evolve and consequences of federalization can thus also vary in time. Therefore, analyzing federal systems over a longer time period can introduce more variation and thus help to better understand which factors play which role. For instance, in Belgium, the recent introduction of separate regional and federal election dates and the consequent formation of incongruent coalitions caused at least a temporary strain on the federal system (Sinardet, 2010). This brings us to another element, which recently received some attention in the general literature on federalism but has not really been connected to the paradox question: the role of party politics. In research on federal reform, regions or other sub state collectivities often tend to be looked at as homogeneous, single actors, disregarding possible instrumentalisation of sub-state identities by parties and differences between them (for recent exceptions, see e.g. Sorens, 2009 and Sinardet 2011). 3

Also, we think refining the main interrogation of the federal paradox yields more nuanced results: we should not just look at the direction (moderation or radicalization) but also at the character and precise evolution of sub-state nationalism. This is why we chose to analyse the federal paradox for Belgium by focusing on the evolution of sub-state nationalism among political parties during the federalization process of the past five decades. We thereby pay particular attention to whether substate nationalist and other parties influenced each other. Considered as sub-state nationalist parties in Belgium are Volksunie/Nieuw-Vlaamse Alliantie (VU/N-VA), Rassemblement Wallon (RW), Fédéralistes démocratiques Francophones (FDF), Spirit, Vlaams Blok/Vlaams Belang 1. Salience, positions and arguments in electoral manifestoes Our study is based on analysis of electoral manifestoes. We are however aware that these do not reflect the entire character of a political party, not even at election time. During an electoral campaign a party needs to react to current events, which might result in another distribution of salience in an electoral campaign than in the electoral manifesto. In addition, it wouldn t come as a surprise to see parties or candidates take more extreme positions in electoral propaganda or televised debates than those written in their electoral manifesto, as these are compromises of different factions within a party. Nevertheless we believe that electoral manifestoes are the most comprehensive available expressions of salience, policy preference and reasoning of a political party (Budge, 1987; Volkens, 2002). Salience of the decentralisation issue has been the subject of research in the past. When sub-state nationalist parties win elections, mainstream parties adapt their attributed salience as well as positions, whereas peripheral parties have a more stable degree of attention (Mazzoleni, 2009). Dandoy (2010), by looking at the influence of electoral performance of sub-state nationalist parties on attention to territorial reforms, confirms this contamination hypothesis. Although Mazzoleni (2009 : 212) states that, as suggested by salience theory (Budge et al, 1987), high degree of emphasis on decentralization is related to a positive discourse on decentralizing reforms, we do not believe such a correlation between high salience and positive discourse can always be taken for granted. Moreover, even if we did it wouldn t give us exact information about the nature of these party positions. Therefore we believe research on attention to territorial reform is not sufficient to grasp the full scale of sub-state nationalist positions. In Belgium this also includes references to institutional reform, negative presentation of the other community, positive presentation of the own community, references to conflict between communities or interpretation of conflict in community terms. Changes in party positions have been far less studied than changes in salience. Some authors have argued that institutional reform will also increase demands for autonomy (Rocher et al, 2001) and that over time we observe a convergence in support among mainstream parties for political decentralization in France, Britain, Italy and Spain (Mazzoleni, 2009). 1 We realize that the literature tends to designate Vlaams Belang mostly as an extreme right party, but the party itself clearly defines itself as (sub state) nationalist. Sub state nationalism was also the main reason the Vlaams Blok was created, as a radical desertion from the Volksunie, after the latter had compromised with francophone parties on community issues in the Egmont pact. 4

We are not only interested in what position parties take concerningcommunity issues, we are also interested in the reasoning behind these positions. What type of arguments do parties put forward to defend institutional reform? Can these be considered as typical sub state nationalist arguments or do they rather advocate reform from other perspectives? Do we see an influence from sub-state nationalist parties on other parties or the other way around? And do these arguments change as the federalization process moves further? Similar to the problems encountered by Mudde (2000) in his work on electoral manifestoes of the extreme right, we can t be sure that the party positions advanced by parties are what they really think or whether this use is strategic. However, this is also true for issue salience and positions. Moreover, even if parties would use (or not use) a number of arguments for strategic reasons, can in itself be very revealing of evolutions of a party and larger of a political and societal climate. We divided the arguments into identity-based and functional arguments, whereby a spread of the proportion of the former can be considered as an indication of increasing sub-state nationalism. Identity-related arguments include arguments whereby homogenous, undifferentiated groups with specific collective interests, opinions and identities are at the origin of the argument to transfer the sovereignty of a competence to this group. This type of arguments can be considered as typically nationalist in the sense that they primarily emphasise national uniqueness and intra-national uniformity but largely ignore intra-national differences (Wodak et al, 2009 : 4; see also Martin, 1995). Functional arguments, are those who do not entail such elements, but rather base demands for decentralisation on other types of analysis of governance and society. Belgian federalism Belgium is of particular relevance to the study of federalism and nationalism in general and of the federal paradox in particular, as since the 1960s the unitary Belgian state has been subject to a process of devolution that eventually led it to officially become a federal state in 1993. Since 2007, it has been trying to negotiate a sixth reform, which has so far only resulted in Belgium breaking the world record of the longest government formation and being in a state of permanent political crisis, at least since 2010, when the Flemish-nationalist and separatist N-VA became the largest party in the north of the country. The institutional landscape that resulted from federal reform is very complex. Two types of federated entities were created: three territorially based regions (the Flemish, Walloon and Brussels region) and three language based communities (the Flemish, French-speaking and German-speaking community). The Flemish and French-speaking community overlap in the Brussels region, that is part of both communities. The borders of regions and communities have been based on those of the four language areas, through which language use is officially regulated: only the official language(s) can be used in administration, education, justice. Since 1963 the borders of these language areas have been fixed, through a linguistic border line. The system is thus based on territorial unilingualism (except in the Brussels region). However, some exceptions exist: 16 communes (of which 6 bordering the Brussels region) with significant linguistic minorities enjoy language facilities which grant inhabitants the right to communicate with the authorities or have primary school organised in another than the official language (Sinardet 2008b). 5

The institutions of federal Belgium are both a product and a pace maker of (political) identity construction: they created permanent boundaries that gave additional subjective meaning to cultural markers and / or territory in addition to favouring identity politics (Lecours 2001: 63). Notwithstanding the existence of three regions, three communities and four language areas, the federal dynamic in Belgium is largely bipolar, based on the two large communities of Dutch-speakers (approx. 6 million) and French-speakers (approx. 4 million). On the level of federal parliament and government, a number of consociational devices, obliging power-sharing, were introduced in 1970: all MPs have to belong to either the Dutch or French language group, a number of special majority laws can only be passed by a majority in both language groups (and an overall majority of two thirds), an alarm bell procedure protects one language group from being dominated by the other, linguistic parity is guaranteed in the council of ministers (this is the federal government with the exception of secretaries of state) which also decides in consensus (Deschouwer, 2006; Sinardet, 2010). On the level of the party system, Belgium is also a unique federation, as no national parties of importance exist, the three traditional parties having split on language basis between 1968 and 1978 and new parties created afterwards (greens, extreme right, ) limiting their action radius to one language community (Deschouwer, 2009). On the level of the electoral system, for the Senate and the European Parliament two electoral colleges were created and for the Chamber most electoral districts do not cross the borders of the regions. This all leads to federal elections being in fact community elections : community parties compete with parties of the same community for community voters through community campaigns. After election day, however, two community election results are put together to form one federal government (Sinardet, 2008). Although the Belgian system and practice of consociationalism and federalism was supposed to lead, amongst other objectives, to political pacification between the communities (Deschouwer 2002), the bipolar institutional characteristics are rather incentives that foster political conflict. Due to the way the party and electoral system is organized, parties only compete for votes within their own language community and are therefore not incited to take into account or be accountable to voters of the other language group. This stimulates taking polarized positions on community issues and also leads to other issues being framed in a community dimension (cf. De Winter, 1993; Sinardet 2010). Institutional explanations can therefore largely account for community conflicts being much more salient among political elites than among the Belgian population (Hooghe 2004). Indeed, public opinion research shows that community issues generally score among the lowest as votedetermining issues among Dutch-speaking as well as French-speaking voters (Deschouwer & Sinardet, 2010). Also the number of separatists remains limited to 9,4 % in the Flemish region and 3,84 % in the Walloon region (Swyngedouw and Rink, 2008; Frognier et al, 2008). Research on ethnoterritorial identity feelings shows a majority of citizens still identifies with Belgium and does not at all consider Flemish or Walloon / Francophone identity on the one hand, and Belgian identity on the other to be mutually exclusive (Deschouwer & Sinardet, 2010). Data and Methods 6

As mentioned, our corpus consists of the electoral manifestoes of all parties that got representatives elected in parliament during national / federal elections from 1965 to 2010 2. When regional electoral manifestoes were mixed with or attached to the federal electoral manifesto, these were also coded 3. This totals 150 electoral manifestoes of 16 parties. In the changing ecology we tried to construct a coherent evolution of political parties. Which parties were considered a successor to each other as well as the abbreviations used in this paper can be found in annex 1. Measuring salience, positions and arguments Electoral manifestoes have been studied before, but mostly quantitatively such as in the Comparative Manifesto Project (Budge et al., 1987), while we want to combine a more quantitative aspect (salience of community issues issues) with a qualitative focus such as party positions and argumentation. The combination of quantitative and qualitative methods allows to discover nonobvious relations and gives more room for interpretation of the data. Although through a quantitative method such as corpus linguistics the researcher can establish more objective (Baker et al., 2008) results, it can also lead to misinterpretation. An illustrative example is the case of Freake et al (2011) who measured discursive construction in Québec and noticed that the computer program wordsmith misinterpreted references to our nation from a first-nation inhabitant as references to Québec. Also, even for our analysis of salience, the data gathered by the comparative manifesto project, although highly valuable, are less useful, because the only codes concerning community issues are decentralization positive and decentralization negative, which is far too limited for our purposes. In the Comparative Manifesto Data decentralization was defined as support for federalism or devolution; more regional autonomy for policy or economy; support for keeping up local and regional customs and symbols; favourable mentions of special consideration for local areas; deference to local expertise (Volkens & Würst, 2003 : 26). Dandoy (2010) expanded this concept to include inter-level relations, state reform, specific status of Brussels and linguistic issues. However, because we find this definition doesn t contain the full extent of community issues in Belgium, we created the concept of community issues and established a clear delimitation of what we would code under this (see table 1). Also, we think salience should not only be measured through the proportion of the program that is dedicated to community issues, as parties can also express the importance of an issue through the prominence they attribute to it in an electoral manifesto (an issue discussed on the first page, will be perceived as more salient than those discussed last). We have therefore also added this dimension in our salience index. 2 With the exception of the Communist party, of which programs were not available. The communist party should have been included in 1965, 1968, 1971, 1974, 1977, 1978 and 1981, 3 This was the case in 1971, when the liberal party, had a separate Flemish and Walloon section, similar the PSC and the CVP had the same program in 1971, except for a regional part. In 1974, again the CVP and PSC have the same program, except for a regional chapter. the PSC has a regional (Walloon) section. In 1974, the PSC and CVP have the same program, except for a regional section. In 1977, the BSP has a second part of its program called regional policies, divided in a part on Wallonia, Flanders and Brussels. In 1985, the PSC has a separate program for Brussels. In 1995, the PS has specific programs for Wallonia, Brussels and the German community, the PSC has a separate program titled regional elections 1995. ln 1999, the PRL-FDF-MCC formation has chapter on Wallonia-Brussels between a chapter on Justice and a chapter on Europe, the Flemish Green party indicated with (VL) or (B) or (EUR) which propositions belong to which level of government. 7

Table 1 : definition of community issues community issues references advocating institutional reform. arguing in favour or against federalism tables, cartoons, numbers showing discrimination by the central state, if dominated by one community. negative references to the other community positive references to the own community as a distinct culture and/or identity References to conflict between communities or interpretation of conflict in community terms : language quota s, Brussels, periphery around Brussels, removal of French-speaking University of Leuven, and Voeren. not community issues decentralization to provinces or municipalities institutional reform concerning institutions such as parliament, government, senate and the administration institutional reform within Dutchspeaking community or French-speaking community such as cooperation between Brussels and Walloon region. The left column of table 1 contains what we conceive to belong to community issuesin Belgium, while in the right column are institutional reforms that aren t necessarily associated with community issues in Belgium. A definition of community issues would be all text that includes all normative references to differences, all references to tensions between different regions and communities and normative expressions concerning the own or another region/community or devolvement of competences to regions/communities. Subsequently we created an index of the salience of community issues in electoral manifestoes, which took into account the percentage of the total manifesto devoted to community issues, similar to what was done in other research, as well as the prominence attributed to community issues in the manifesto. To calculate this salience index, we coded every page of every electoral manifesto, by giving it a value from 0.1 to 1 depending on how much of the page was devoted to community issues. We admit that this is rather rough, but this allowed us to deal with quite a large corpus, while treating small as well as large electoral manifestoes in a consistent manner. This consistency is not to be taken for granted, as for example in the Manifesto Project sometimes a paragraph was used as a coding unit instead of the habitual quasi-sentence (Klingemann, 1987 : 301). We then calculated the percentage of the electoral manifesto which is devoted to community issues. Then we multiplied the proportion of the 8

electoral manifesto dedicated to community issues by the prominence index. This prominence index was calculated by dividing the number of the first page on which community issues appeared by the total amount of pages. Subsequently we subtracted the outcome from 1. This way, the higher the amount of the manifesto that is devoted to community issues and the closer to the first page, the higher the salience index will be. An increasing salience index points to a rise of sub-state nationalism, while a decreasing salience index points to a decrease in sub-state nationalism. To analyse party positions, we also couldn t rely on the CMP. Indeed, as mentioned by other authors (Janda et al, 1995 : 177), the data are not very useful to determine parties actual positions on issues. We used 320 codes to code positions on community issues in consecutive elections. These codes thus present party positions such as expansion of Brussels outside 19 municipalities or devolution of national lottery, we coded them as 1 if a political party mentioned favoured the position and as -1 if they opposed the position, we left a blank when parties didn t mention the position. After an initial reading of party positions advocated in the electoral manifestos, a list of all party positions related to community issues was drawn up. This list was tested and adapted after a second reading. In a third phase, a definitive list was created, after which each electoral program was coded. In the last phase, a final reading of all electoral programs was done to track possible positions that were missed, resulting in a complete coding of all positions on community issues of all parties represented in the Belgian parliament between 1965 and 2010. Concerning the analysis of arguments, we used an inductive, grounded approach (Glaser & Strauss, 1971; Strauss & Corbin, 1990) in the sense that we didn t deduct arguments used to devolve competences from theory or approach the data with a start list in order to avoid forcing codes onto the data. The unit of analysis is an argument, this is one or several sentences in which a position is defended. One sentence can contain several arguments and thus be categorized under different arguments. We started by clearly limiting our research focus to arguments on devolution of competences to regions and communities. After an initial reading we drew up a list of arguments used in the electoral manifestoes. In the process, some categories were merged and others were split. Subsequently a second reading followed, after which the same splitting and merging process took place and led to a definitive list. In the third phase, all electoral manifestoes were coded with this list. In the last phase, the electoral manifestoes were coded again to make sure we coded all arguments. This list of arguments found consists of 18 different categories, which are summarized in table 2. First, we discuss the identity-based arguments. The first is a people needs to rule itself. Although political parties only rarely use this argument to argue in favour of total independence, the logic that there is a distinct group of people, which needs to rule itself at least on a number of important domains is also part of this category. The second is Belgium is ruled by the other language group, which is defined as the other language group is in control of the Belgian government, or Belgium is ruled in function of the other language group, discriminating against our language group. The idea of internal colonialism and an expansion and dominance of the centre of the periphery (Deutsch, 1966; Hechter, 1975) was seen as an explanation for the rise of sub-state nationalist parties, although it has been discredited in by more recent studies (Swenden, 2006). The third is x is limited by Belgium/other language group, defined as Belgium or language group B hinders/restrains/limits language group A. The other is not perceived as a actively discriminating subject, but rather a 9

passive, hindering object. The fourth is our own money in our own hands, referring to quotes stating that a community should be able to keep its money, it should be able to dispose freely of its own resources. The fifth is the region is a better manager, meaning that devolution will lead to better policy, because the community/region is better in administering this competence. This argument is based on identity, because it supposes a region or even group which would inherently be better at managing a certain competence. The sixth is structural differences and includes that regions/communities should be in control of a competence because they have structurally different characteristics, needs, problems which warrant a specific policy on the regional level. The seventh is cultural differences, which are all the arguments stating that the other region has a different culture, attitude, policy opinions therefore a competence needs to devolved. Next, we discuss the functional arguments. The first is national pacification, which means that a devolution of competences is needed in order to pacify community conflicts. The second is better cooperation between regions/communities, which argues that a devolution of competences will result in better cooperation between regions/communities. The third is more prosperity, which includes all arguments that argue that a devolution of competences will result in more prosperity for the whole or for the region. Often this is expressed by arguing that more autonomy will result in economic growth. The fourth is more efficiency, meaning arguments that argue that devolution will result in more efficiency, it will be done quicker or cheaper, because supposedly less civil servants would be necessary. The fifth is more responsabilisation, which states that devolution will lead to regional governments being more responsible for the financial repercussions of their policies. This is supposed to lead to a more cautious and efficient usage of budgetary means. The sixth is more coherence, which means that devolution is necessary to obtain more coherent and/or more homogeneous policy packages. The seventh is more modernisation. This code consists of the arguments that devolution of competences will lead to a modernization of the government. The eight is more democratic, which argues that devolution of competences makes the government more democratic, brings the government closer to the people or raises democratic legitimacy. The ninth is simplification, arguing that a devolution of competences will lead to a simplification of the institutions and/or more transparency. The tenth is external legitimation. This code contains all arguments that legitimize a devolution of competences because it is advocated by external actors such as labour unions or experts or because it works well in other countries or is in line with the evolution of the European Union or if it s the will of the people. Last, we included a category unfinished, which we didn t include in the list functional arguments. Arguments in this code state that the institutional reform is natural evolution, unfinished, impossible to avoid. Table 2 : identity-based and functional arguments Identity-based Functional Other People rule itself National pacification Unfinished Ruled by other More cooperation Different culture More efficiency Own money in own hands More responsabilisation Hindered by other More coherent competences Structural differences Modernization Better policy Democratization 10

Simplification External legitimation To estimate the proportion of identity-based arguments and functional arguments, we divided the sum of identity-based arguments with the sum of functional arguments. We have called this the identity index. When this identity index rises, this means that there is a rise in sub-state nationalism and vice versa 4. Our data only extends to 1965, which allows to observe evolutions in these arguments, but it is not totally excluded that arguments that we designate as having originated from sub-state nationalist parties have in fact previously been used by other parties. However, due to time constraints we were obliged to cut off our data at some point. Results Salience of sub-state nationalism When we take a first general look at the total of election manifestoes, we can clearly see that the community issue remains present over time. Out of 150 analysed manifestoes, only 4 have absolutely no attention for sub-state nationalism. This observation might however partly be explained by the size of the Belgian electoral manifestoes, which tend to cover a wide range of issues. Figure 1 : Average salience index for Dutch-speaking mainstream parties (window) and Dutchspeaking sub-state nationalist parties (square) 40,00 35,00 30,00 25,00 20,00 15,00 10,00 5,00 0,00 1965 1968 1971 1974 1977 1978 1981 1985 1987 1991 1995 1999 2003 2007 2010 4 It happened that parties didn t use an identity-based argument, this would mean that in the calculation of the identity index we would have zero in the numerator. Or, similarly it happened that an electoral manifesto didn t use any functional arguments, thus we would have zero in the denominator. To avoid this situation we added 1 to all sums of arguments. 11

Figure 2 : Average salience index for French-speaking mainstream parties (window) and Frenchspeaking sub-state nationalist parties (square) 40,00 35,00 30,00 25,00 20,00 15,00 10,00 5,00 0,00 1965 1968 1971 1974 1977 1978 1981 1985 1987 1991 1995 1999 2003 2007 2010 For the Belgian case, which consists of two separate party systems (Billiet et al., 2006; Sinardet, 2008; Brack & Pilet, 2009), it is useful to look distinctly at the Dutch-speaking and French-speaking party system in order to analyse the dynamic between sub state nationalist and other parties. Therefore, in figure 1 we have separated the Dutch-speaking mainstream parties from the Dutch-speaking substate nationalist parties, while in figure 2 we did the same for French-speaking parties. As can be expected, sub state nationalist parties attribute consistently higher salience to community issues than other parties. However, when we take into account the entire analysed period, we see a diminishing salience over time for Dutch-speaking sub state nationalist parties. Surprisingly, in the 2010 election, which was strongly dominated by community issues, the index for N-VA and Vlaams Belang is at an all time low, for VB with the exception of 1995. Here, we do however notice a difference between the prominence and proportion index, as prominence decreases for sub-state nationalist parties since 1999, but proportion increases from the all-time low in 1995. On the long term, it is also not possible to speak of an increase for other parties, even if in 2010 the index is suddenly at its highest point since 1987. For the French-speaking parties, the image is less clear, which is also due to the fact that independent sub state nationalist parties only existed until 1991. There is however a clear increase in salience from 1978 to 1985, with a remarkable peak in 1981, 12

after which figures remain very stable, be it at a slightly higher level than for the Dutch-speaking parties. After 1981, the prominence index rises to a peak in 1985, however the proportion index is much lower. Actually, over the entire period, salience of community issues is structurally higher among French-speaking than among Dutch-speaking mainstream parties, which is quite surprising, given the fact that sub state nationalism is generally mostly attributed to Flemish parties. It seems that salience of community issues is past its peak for French-speaking parties since 1981 and for Dutch-speaking parties since 1987, although for Dutch-speaking parties it has risen in 2010, but is still lower than for the French-speaking parties. To summarize, on the basis of these figures, it is difficult to speak of a clear, long term increase in salience of community issues. Nevertheless, if we look in more detail at evolutions per election on figure 1, we do observe that a peak in salience by Dutch-speaking sub-state nationalist parties is often followed in the subsequent election by a peak in salience by Dutch-speaking mainstream parties, for example in 1965-1968, 1977-1978, 1985-1987.In addition, the proportion that Dutch sub-state nationalist parties attribute to community issues peaks in 1977 and 1985, after a low level in 1981. After 1985, the proportion index drops until a low in 1995. While the prominence index is higher than average in 1977 and 1985, it only decreases a little and even peaks in 1995, when the proportion is low. For French-speaking parties, on figure 2, the peak in salience of the French-speaking mainstream parties, coincides with a peak in salience of the sub-state nationalist parties in the same election in 1968 and 1981. Concerning the French-speaking mainstream parties it is remarkable that after the disappearance of the last independent sub state nationalist party, the FDF, the salience index remains very stable, although not at the lowest level. Both observations point to an adaptation of parties to the presence of sub-state nationalist parties, concerning salience. To be able to look at the dynamics between the different parties more in detail, we have included the salience indexes for every individual party per individual election year in figure 3. As can be observed from the table in figure 3, the higher index for French-speaking mainstream parties between 1971 and 1991 can largely be attributed to the rise in score of the French-speaking socialists, which is consistently higher in that period than that of the other francophone parties. The 1981 peak can almost entirely be explained by the fact that the PS suddenly devotes almost half of its manifesto to community issues, the highest score for any party. In that 1981 manifesto, the PS wants to regionalize the five so-called national economic sectors, as it feels that on the federal level Dutchspeaking parties and particularly the CVP refuse to aid these ailing economic sectors (PS, 1981 : 21). The drop in the salience of French-speaking mainstream parties after 1991 is again most attributable to the PS. Possibly, this is the result of the disappearance of the FDF as an independent party after 1991. When the RW disappeared after 1981, a drop in salience for French-speaking mainstream parties could also be observed. One could also look at the possible causality in this relation the other way around, as wen the salience index of the French-speaking socialist party starkly peaks in 1981, the Walloon sub-state nationalist party contests its last election and when that the French-speaking liberal party peaks in 1991, the Brussels sub-state nationalist party contests its last election. 13

Figure 3 : salience index per political party, per electoral year. 60 50 40 30 20 10 0 1965 1968 1971 1974 1977 1978 1981 1985 1987 1991 1995 1999 2003 2007 2010 VU/NVA 20,8 18,77 14,43 16,88 35,14 27,5 13,14 41,18 38,28 14,22 18,31 8,85 11,11 23,28 2,73 FDF 33,83 20,55 11,62 28,65 27,55 27,5 18,13 26,31 18,08 RW 14,29 3,64 23,61 26,47 30,1 VB 22,22 10,94 22,97 13,47 21,25 5,96 22,56 20,95 16,61 16,58 Spirit 4,92 0,28 SP 0,44 15,2 4,88 0,28 0,4 0,28 0,1 1,35 0,83 0,37 0,15 0,01 0 0 0,07 PS 0,44 15,2 4,88 0,28 0,4 13,59 49,6 11,81 8,42 6,46 1,1 2,06 1,41 1,18 0,19 VLD 0,66 12,58 1,78 2,34 0,35 2,26 0 0,83 8,5 1,2 0,86 0,64 0,61 1,07 6,36 MR 0,66 12,58 0 7,27 1,51 5,86 1,97 3,63 3,23 7,35 0,86 0,45 2,65 0,17 3,03 CVP 0,74 1,78 0,74 1,3 0,59 2,5 0,62 0,41 0,63 2,89 0,2 1,23 0,46 0,29 1,82 cdh 0,74 8,89 1,21 0,69 0,62 5,13 6,95 5,65 2,77 3,38 0,73 1,09 0,07 0,41 0,96 FN 0,58 0,38 Groen 0 0,25 0,14 0,03 0,32 1,02 0,37 0,25 0,67 Ecolo 1,54 0,26 1,22 0,34 0,08 0,47 0,52 0,34 2,45 LDD 2,21 2,22 VU/NVA FDF RW VB Spirit SP PS VLD MR CVP cdh FN Groen Ecolo LDD Within the Dutch-speaking party system, differences between parties are less striking. The relatively small peak for Dutch-mainstream parties in 1987 can however be attributed to the liberal party, which also contributes to to the 1968 peak, together with the SP. These two cases however do not seem to refer to a more structural pattern such as with the PS. Among Dutch-speaking parties, the CVP has the most consistent and low salience index, with all figures contained between 0,2 and 2,5. We will now refine these findings based on saliency and see if they concur with those on the policy positions. 14

Policy positions When we look at the positions taken in the total of electoral manifestoes, a first surprising observation is that in 112 out of 150 of these, more autonomy for the regions/communities is advocated. Even parties which are considered as being rather moderate on community issues often argue for devolution of at least some competences. For the Dutch-speaking socialist party this is the case in 11 out of 15 manifestoes, while for the Dutch-speaking Greens this is four out of nine. In what follows we will more specifically look at the evolution of federal reform positions concerning state structure, social policy, foreign policy and financial/fiscal autonomy and describe an evolution common to these policy domains. State structure State structure refers to the type of state the parties would like to see Belgium become and entails unitary (1), decentralized unitary (2), federal (3), confederal (4) and abolished (5). Rather than determining parties positions on state structure on the basis of our own definition of these categories, we coded based on the preferences parties themselves explicitly mentioned in their electoral manifesto. This is because there is at times a sensible difference between the definition parties seem to give to types of state structure and definitions on which there is a broad international academic consensus 5. Even if labels are not used correctly, the fact that they are used is not without importance as it does signal a fundamental positioning in the debate. In figures 4 and 5 we see an overall evolution towards a less unified state structure in the period from 1965 onwards. While in 1965, all mainstream parties still favoured a unitary state structure, with only the still unitary socialist party arguing for decentralisation within this framework), in 1974 a general acceptance of some form of decentralization to communities or regions had already occurred, which turned into a general demand for federalism in 1991 all parties favoured federalism (which would be inscribed in the constitution two years later), when the Dutch-speaking socialists and the Francophone liberals were the last to join the consensus. In 2010, two non-sub state nationalist parties favour confederalism. In this evolution, other parties clearly followed sub-state nationalist parties position on state structure, first in a consensus towards decentralization, then federalism and for Dutch-speaking Christian-Democrats and liberals further to confederalism. Sub state nationalist parties were always the first to advocate a new type of state structure (with the exception of decentralisation within a unitary framework). Thus not only mainstream parties radicalize, but sub-state nationalist parties as well, with most recently the confederalist VU re-emerging as the separatist N-VA. 5 For instance, from 2003 onwards, some Flemish parties plead for a confederal state structure, while their concrete positions would rather be associated with a federal system, be it with substantially more regional autonomy than before. 15

Figure 4 : state structure position Dutch-speaking parties SP VLD CVP Agalev VU VB 1965 1968 1971 1974 1977 1978 1981 1985 1987 1991 1995 1999 2003 2007 2010 Figure 5 : State structure position French-speaking parties PS MR cdh Ecolo FDF RW 1965 1968 1971 1974 1977 1978 1981 1985 1987 1991 1995 1999 2003 2007 2010 This evolution towards a more decentralised state structure is at first the same among Dutchspeaking as well as French-speaking parties, but among Francophone parties this stops in the 1990s, as they stick with a federal organization of the Belgian state, arguing that institutional reforms have brought our country to a finished federalism. We now have to let these institutions function at the heart of the Belgian federation in respect for federal loyalty (PS, 1995 : 89) and that an institutional equilibrium (Ecolo, 1999 : 1; PSC, 1999 : 1; PRL-FDF, 1999 : 9 ) has been reached. Simultaneously, further radicalisation concerning state structure occurs on the centre-right of the Dutch-speaking political spectrum, with the Dutch-speaking Christian-democrats and later also the Dutch-speaking liberals advocating confederalism (even if their further positions on state reform would not be 16

considered as confederal, based on usual definitions). During a 2001 party conference, the Dutchspeaking Christian-Democrats chose to change their preferred institutional organization to confederalism, while the Dutch-speaking liberals followed this choice in their 2010 electoral manifesto. Thus on the Dutch-speaking side the parties ideologically closest to the sub-state nationalist party seem to follow the sub-state nationalist party the furthest, as Flemish nationalism has always been more of a (centre-)right political movement. This observation can however also be made for francophone parties. Particularly illustrating in this respect is the behaviour of the French-speaking liberal party. Among mainstream francophone parties, it is the first which moves in 1974 to federalism, while in 1976 two of the leading figures of the Walloon sub-state nationalist party join its ranks, after which the RW goes back to a more left-wing course (Deschouwer, 2009). After the left turn of the RW, the French-speaking liberals change their preference to decentralization in 1978, and the PS change their party position from decentralization to federalism. This suggests an influence of sub-state nationalist parties on those ideologically close to them for the French-speaking party system also. We discuss state structure, because of its importance, but also because it shows a similar evolution as is the case for the competences of foreign policy, social policy, financial/fiscal autonomy. Broadly, in the Dutch-speaking party landscape the sub-state nationalist VU proposes devolution of a certain competence, it is followed by the BSP, later on by CD&V, then VLD. When the BSP splits the Dutchspeaking socialists take less sub-state nationalist positions, than the French-speaking socialist party does. The Dutch-speaking left-wing parties don t adopt a sub-state nationalist position often, but seem to do so more often since the nineties. In the French-speaking party landscape we observe that a position by FDF or RW, taken up by PS, then cdh and last and hesitantly MR. A radical shift occurs for all French-speaking parties in after 1991 when they start opposing all further devolution. Social policy, fiscal autonomy and foreign policy We consistently observe a similar evolution in other policy domains such as social policy, fiscal/financial autonomy and foreign policy, where the Dutch-speaking Christian-Democrats and the French-speaking socialist party (and before the split of the socialist party this is true for the Belgian socialist party also) are the parties that are the first and go the furthest in adopting sub-state nationalist party positions. Concerning foreign policy, the VU is the first to advocate more autonomy in cultural international relations in 1971, and is followed by the CVP in 1977. That same year, the BSP is however the first party to demand direct representation in the EC, which is also advocated by the VU in 1985 and by VLD in 1995. As of 1985, the VU also demands autonomy in international relations for the regional and community competences, as well as the possibility to close international treaties. In 1991, the Dutch-speaking socialists follow by demanding the possibility to have international relations, but don t follow on the demand to close treaties. That demand is however followed by the Dutchspeaking Christian-Democrats as of 1991In 2007, the CD&V stresses again the need for an own foreign policy for Flanders and direct representation in Europe. In the French-speaking party landscape we observe that the FDF as well as the PSB demand more autonomy concerning foreign policy and direct representation on the European level in 1977. The 17

only other party to still follow is Ecolo, only in its 1985 manifesto. The PS continues to follow substate nationalist demands of the FDF on foreign policy. When the FDF demands the right for the regions and communities to conclude treaties in 1987, the PS duly follows in 1991. However, from 1995 onwards, devolution of any aspect of foreign policy is entirely absent from the francophone parties manifestoes. The only exception is the demand of the PS to split development cooperation in 2003 and 2007 (which they had already done between 1985 and 1991). However, the three other francophone parties clearly state they are opposed to such defederalisation 6. The story concerning social policy reads quite similarly. The VU is the first to plead for devolutions in social policy, with family policy and public health in 1968, while in 1971, it demands a split in social security as a whole. In 1977, it is followed by the BSP, which demands a split of family policy and public health. On the split of health care, the VU is again first and followed by the CVP in 1991, who continue to demand this in every electoral manifesto onwards. The VLD only follows splitting health care in 2003, although they already advocate splitting parts of social security that don t concern replacement income in 1991. The sp.a and Groen! remain strongly opposed to the defederalization of social security and specifically of income-replacing allowances. However, in 1991, the sp.a does state that the distortions between the communities on the level of health care, which are a.o. related to abuses and oversupply are * + unacceptable and if these imbalances remain unresolved, the federalisation of additional allocations will undoubtedly appear on the agenda. This suggests the party does not demand a split in health care allowances, but feels a strong pressure on this issue, to which it cannot entirely resist. The Greens also start talking about defederalising family policy in 1999.d 1978, while also mentioning social policy from 1971 to 1987. In 1977, the PSB follows on defederalizing public health and family policy, while PRL and PSC also want to defederalize parts of social policy. But while the PRL nuances its position in 1978 and the PSC never reiterates it again, the PS keeps stressing its position on public health and child benefits. In addition, they add in 1985 parts of social policy and in 1991, the devolution of responsibility for the Public Centre for Societal Welfare (OCMW/CPAS). Again, as of 1995, demands on the domain of social policy disappear from all francophone manifestoes. Again, the same dynamic appears concerning financial and fiscal autonomy. The VU is the first to argue in favour of autonomy in spending for the regions in 1968. They are followed by the PSB in 1971, which argues that each level of government will be financially responsible for the competences attributed to it. In 1974 the VU argues that the regions should take their own responsibility to fund economic expansion, while the CVP argues that territory-based taxes should be transferred to the regions and the PVV only asks for enough financial means. In 1977 the PVV however also switches to a well-described autonomy concerning * + finances. From 1985, the VU argues that the regions should be completely fiscally independent and that the federal level should depend on the region. That same year PVV only argues for the competence to grant subsidies and lower taxes, but in 1995, they argue that the two big communities need to cover their expenses by their own taxation, whereas the CVP emphasizes that regional taxes are collected by and controlled by regional authorities. The SP only takes its first position in 1999 and argues for a split-rate income tax, which the regions could fluctuate within margins. The CVP however meanwhile argues for the transfer of the whole income tax and to make it possible to fluctuate the tax on companies. In 2003, 6 The fact that this issue comes to the forefront in 2003 and 2007, is because a partial defederalisation of development cooperation was decided in the 2001 state reform, but had not been executed. 18